A Deep Learning Approach to Glioblastoma Radiogenomic Classification Using Brain MRI

A malignant brain tumor known as a glioblastoma is an extremely life-threatening condition. It has been proven that the existence of a specific genetic sequence in the tumor known as MGMT promoter methylation is a favourable prognostic factor and a sign of how well a patient will respond to chemotherapy. Currently, the only way to identify the presence of the MGMT promoter is to perform a genetic analysis that requires surgical intervention. The development of an accurate method for determining the presence of the MGMT promoter using only MRI would help to reduce the number of surgeries. In this work, we developed a method for glioblastoma classification using just MRI by choosing an appropriate loss function, neural network architecture and ensembling trained models. This problem was successfully solved as part of the “RSNA-MICCAI Brain Tumor Radiogenomic Classification” competition, and the proposed algorithm was included in the top 5% of best solutions.